This project is an Dapp, allowiong users to upload pictures and their comments. Each user has a Metamask account, with an unique address, namely its public key. When a user uploads an image, its similarity with other images which have been uploaded will be checked and the user will be given a warning about the distance(the higher the more similar) between existing similar images.

Once a user login with Metamask account, he/she can start to view posts and upload images. Personal page showing your posts. Press the button and add a new post. Choose a image. Since the information of a post is stored in the blockchain, you have to confirm the transaction and pay the gas. The transaction is successful, and a new post shows up. The image was just uploaded to IPFS Moreover, you can press the camera icon on the navigation bar and take a photo. Also change the filtering mode of the photo.

This Dapp is responsive by device width.

Front end: React, Redux for management of states and events, react-router-dom for front-end routing, Ant Design for UI. Blockchain: Truffle for the developement of smart contracts( Solidity ), Ganache for simulating blockchain locally. Similarity Check: Python Falcon server framework for REST API, Docker for hosting tensorflow server.

• Ganache
• Metamask account note: create a new private network with url: http:https://127.0.0.1:7545
• Install concurrently

    npm install -g concurrently 

1. Download

    git clone https://github.com/Andy-Cheng/picture_blockchain.git
    npm install

2. Open Ganache, and open Metamask, connect to the private network with url: http:https://127.0.0.1:7545 Import an account showing in Ganache, you'll see a new account holding around 100 eth.
3. Compile Contracts and migrate to the blockchain.

    truffle compile 
    truffle migrate (For the first time, latter you have to type truffle migrate --reset)

Note: Every time you migrate the contracts, you have to reset all of your testing accounts on Metamesk. 4. Start up Dapp in localhost:3000 and local ipfs node simultaneously

    npm run dev 

or in one terminal

    ipfs daemon

and in another terminal

    npm start

5. To host servers for similarity checking, pull the image containing our servers and host tensorflow model server.

    docker pull chuanchung/imgsim
    docker run -d --net=host chuanchung/imgsim tensorflow_model_server --port=8500 --model_name="images" --model_base_path="/home/Imgsim-Server/inception_serving" 

Open another terminal and enter the container we've just run.

    docker exec -it [container_id] /bin/bash

Host another server for providing a REST API

    cd /home/Imgsim-Server/
    gunicorn -b 0.0.0.0:8000 --reload img.app